Defect Detection: In-Process Monitoring vs. Non-Destructive Testing

The claim that InnerVoice, or any in-process monitoring capability, can detect defects is a misnomer. In actuality, in-process monitoring infers a defect-free part by directly interogating the process physics and dynamics.

The key term from the previous sentence is “infers”. In-process monitoring draws a conclusion regarding part quality based on the analysis of one or more process signals. The process signals and the corresponding analysis capture the process behavior and represent the evidence from which an in-process monitoring system bases its conclusion. If the process behavior of a production part is similar to the process behavior performed during the qualification effort of acceptable parts (commonly referred to as a "baseline"), then the production part is deemed acceptable, otherwise the production part is flagged as suspect or unacceptable depending upon the severity of the discrepancy.

So, how does in-process monitoring differ from non-destructive testing?  Below are some of the key differences:

• Process vs. Product Attributes: In-process monitoring draws its conclusions from the analysis of process attributes, while non-destructive testing draws its conclusions from the testing and analysis of physical part attributes.
• On- vs. Off-Machine: In-process monitoring is performed on-machine and in (near) real-time, while non-destructive testing is (usually) performed off-machine and post-process. 
• Automated vs. Manual: In-process monitoring is automated and requires no human intervention, while non-destructive testing is typically a manual operation and usually requires human interpretation.
• General vs. Specific: In-process monitoring approaches the need for quality surveillence from a general perspective, while non-destructive testing approaches the need for quality surveillence from a specific perspecitve.

This last bullet item represents an important distinction and requires further explanation. In-process monitoring searches for differences in process behavior without any a priori knowledge about the behavior of interest or what the behavior may suggest. If the process behavior is different from an established baseline of acceptable welds, then an in-process monitoring capability will flag the part and alert the manufacturer of the discrepancy. On the other hand, most non-destructive testing methods search for a specific set of defects or discontinuities. Below are some of the more commonly used non-destructive testing methods and their defect detection capabilities for welding and joining applications:

• Visual inspection involves an operator looking at a test piece and allows for the detection of surface discontinuities such as misalignment, undercut, and corrosion.
• Eddy current testing involves the use of a magnetic field to test for surface and near-surface cracks, lack of fusion, and corrosion.
• Liquid penetrant testing involves the application of a liquid with very low viscosity to expose visible indications of surface flaws such as cracks, porosity, and incomplete fusion. 
• Radiographic testing involves the application of x-rays or gamma rays to detect most any surface or subsurface flaws or defects.
• Ultrasonic testing involves the application of high-frequency sound to detect discontinuities such as inclusions, porosity, and cracks.

As demonstrated in the list above, non-destructive testing methods are limited in their defect detection capabilities. In addition to the types of defects that a particular method may be sensitive to, other limitations may include surface or subsurface capabilities, and the size and orientation of a defect or discontinuity.  In-process monitoring, on the other hand, simply tells the manufacturer whether the behavior is similar to or different from a baseline. Depending upon the in-process monitoring capabilities, further exploration of the data may be available in terms of prognostic capabilities. 

In conclusion, in-process monitoring infers a defect-free part by comparing its process behavior to one or more parts of known acceptable process behavior. Additionally, in-process monitoring will flag a part that requires further quality scrutiny when its process behavior is sufficiently different from this baseline. When a part is flagged by an in-process monitoring system, that part may exhibit behavioral characteristics that are similar to a previous part with a specific defect. In such a case, it would be correct to say the part exhibited behavior indicative of a particular defect. Here, the key phrase is "behavior indicative of a particular defect". However, to say that an in-process monitoring system is capable of detecting a particular defect is, in the strictest technical sense, not true.

While significant differences exist between the two quality surveillence methods, both methods, in-process monitoring and non-destructive testing, are necessary capabilities for a quality-assured process. In-process monitoring is the direct interrogation of process physics and dynamics. Without it, a manufacturer must rely exclusively on machine tool monitoring and post-process, non-destructive testing. In-process monitoring complements these other technologies and provides valuable process information not available through other means.

To learn more about in-process monitoring, see the following resources:

• "What is In-Process Monitoring?" 
• "Introducing InnerVoice"

To learn how an in-process monitoring capability can complement and extend your quality-assured welding process, see the following resources:

• "Creating an Inspect-for-Cause Approach to Manufacturing Quality Assurance through Real-Time, In-Process Monitoring"
• "Eliminating the Blind Spot in a Traditional Approach to Manufacturing Quality Assurance". 

Or contact us today to learn more about how InnerVoice can help migrate your traditional quality assurance approach to an in-process approach.